- Email: [email protected]

Pergamon

0038-1098(95)0056&4 WEAK ANTIFERROMAGNETISM

Vol. 96, No. 8, pp. 535-537, 1995 Elsevier Science Ltd Printed in Great Britain 0038-1098/95 $9.50+.00

IN YFeOs AND HoFeOJ

D. G. Georgiev, K. A. Krezhovt

and V V Nietz

Frank Laboratory of Neutron Physics, Joint Institute of Nuclear Research , 14 1980 Dubna, Russia tInstitute for Nuclear Research and Nuclear Energy, Bulgarian Academy of Sciences, 72 Tzarigradsko Chaussee Boulevard, 1784 Sofia, Bulgaria

(Received 19 January 1995; accepted in revised form 9 August 1995 by P. B&et)

We have performed neutron dimaction measurements of the weak antiferromagnetic component of the noncollinear spin arrangement of Fe3’ ions in YFeGs and HoFeD by means of the SNIM-2 neutron spectrometer at the IBR-2 pulsed reactor of JINR, Dubna. Single crystals were used. The ratio of the week antiferromagnetic component & to the basic antiferromagnetic component G. was found b/G, = 0.014 for YFeOs and b/G, = 0.011 for HoFeOs within error limits of about 10 %. Keywords: A. magnetically ordered materials, C. crystal structure and symmetry, D. exchange and superexchange, E. neutron scattering

contributions it was employed the constant wavelength diffraction technique with polarized neutrons and analysis of spin reversal. The ratio f&,/G. had been estimated equal to (1.93 f 0.18) 10m2which is to be compared with the predictions of the theory’, 4, ’ Ab/ G, = 1.95 . 10e2 in the case of YFeOs. Although the careful additional measurements which had allowed to eliminate the effects of multiple neutron scattering do seem convincing it was found necessary to normalize the peak intensities associated with ,& against integrated intensity of other appropriate nuclear and magnetic peaks which had been strongly affected by extinction. But, it is well known that large extinction corrections cast certain doubts on the final result. Therefore, despite the practically coincident theoretical and experimental estimates of the quantityAb we believe that carrying out measuremeats _ free of such kind of normalization will be of interest.

The crystalline symmetry of YFeOs and HoFeOs is described by the orthorbombic space group Dlt--Pbnm. In both YFeGs below TN x 640K and HoFeG3 within the range of temperatures from 640 K to 63 K the noncollinear antiferromagnetic state Fd (G,F&,) is established. In this spin arrangement the basic vector of antiferromagnetism G = I$ - 22 + A& - $4 (A4i denotes the magnetization of the i-th iron sublattice) is directed along a-axis and there is a small admixture of the vector of weak ferromagnetism 3 = &_+&_+& +& and the vector of weak antiferromagnetism A = 441 - A& - & + A& which point along c- and bcrystal axes, respectively. The presence of the constituents F, and Ab is permitted by crystal symmetry arguments and their magnitudes are determined mainly by the antisymmetric exchange interaction between the Fe3+ ions’, which is about two orders of magnitude lower than the isotropic exchange interaction. However, in contrast to the component F, which can be easily determined the measurement of the weak antiferromagnetic component Ab and hence determination of the associated antisymmetric exchange constant are inaccessible to the common magnetic methods.

The measurements of the component Ab in the present study were carried out on single crystals of YFeG3 (3 mm thick plate, weight 0.3 g) and HoFeOs (dimensions 5 x 6 x 8 mm) by means of the neutron time-of-flight technique using the spectrometer SNIM-2 at the pulsed reactor IBR-2 at JINR6. The simple experimental geometry has been used: reactor core with neutron moderator-bent Ni-coated mirror neutron guid+sample at a distance of 30m from the moderator-single neutron detector located at 2.17 m from the sample. The solid angle of the detector-a cylindrical 3He-filled neutron counter (45 mm diameter, 20 mm thickness of the in-beam sensitive volume) ensured the collection of all neutrons undergoing diffraction on the sample. In order to avoid the need of calibrating the weak diffraction peak

A tool sensitive enough for determination of Ab is the neutron diffraction and such measurements have been performed already2. However, in the cited study it had been necessary to overcome the serious difficulties arising from the relatively rather small scattered intensity attributed to the magnetic component Ab. In these measurements performed on a single crystal of YFeGs about 4mm thick in order to separate the weak diffracted intensity on the background of strong nuclear and magnetic coherent scattering 535

WEAK ANTIFERROMAGNETISM

536

IN YFeO, AND HoFeO,

Vol. 96. No. 8

bility of multiple scattering) was placed at the sample position and prolonged time-of-flight measurements were carried out using the same neutron detector until a sufficiently high counting statistics was accumulated.

_

YFe0,(201) t=64ps

OT

I 480

The single diffraction peak (201) associated with the weak antiferromagnetic component Ab only was measured. Fig. 1 shows the neutron diffraction profile of (201) reflection of YFeOr measured at two different scattering angles 28 = 90” and 94”. In determining the (201) peak area the background was approximated by a straight line. Fig. 2 illustrates the corresponding scattering diagram (at 28 = 90”) and the reciprocal lattice sections with the Ewald sphere.

290 K 15h

I !m

I 520

Channel number

Figure 1. Neutron diffraction patterns for YFeOr at the two given scattering angles; t is the channel width of time analyzer.

intensity against other peaks affected by extinction we first measured carefully the wavelength dependent intensity and shape of the incident neutron beam and the diffracted intensities were further normalized against the primary beam spectrum. To do this a sample of vanadium (a stack of four metallic cylinders with a diameter of 8 mm and 10 mm high each, separated by thin plates of Cd to diminish the proba-

As it is shown the equatorial section of the Ewald sphere is smeared reflecting the assumed angular spread of the primary beam in the horizontal plane within 0.5”. The actual distribution width (at half of the maximum height) at the neutronguide exit is 10’ for neutrons with a wavelength of 3.5 A corresponding to the Bragg reflection from (201) crystalline plane, with neutron-guide transmission featuring angular distributions without extended tails. The Ewald sphere section with the next neighbouring layer of the reciprocal crystal lattice is also shown. Here, the point (104) located near the Ewald sphere corresponds to the components of the spin arrangement which are absent in the magnetic phase f+ On the other hand the lattice point (203) similarly to the point (201) is associated just with a small component A and is unable therefore to affect appreciably the (201) reflection. A similar Ewald sphere construction for the case 28 = 94” shows that again there are no reciprocal lattice points on the Ewald sphere which could influence the (201) reflection. Therefore, the authors firmly believe that the effect of multiple scattering is avoided and the results are not affected by diffraction contributions originating from reflections on crystallographic planes other than (201).

io3, io2,

io1,

-ll

Figure 2. Scattering diagram at 28 = 90” and Ewald construction for YFeGr.

720

740

760 7m am Channel number

U!D

I

Figure 3. Neutron diffraction profile of (201) reflection of HoFer. The spectrum of the crystal rotated by 1S” from the diffraction position is labelled by triangles.

Vol. 96, No. 8

WEAK ANTJFBRROMAGNBTISM

The treatment of the neutron data on YFeQ for (20 1) reflection results in the following magnitude of the component of the effective spin S (per one Fest-ion) directed along baxis: Sb(28 = 90”) = 0.0337 and S,([email protected]!3= 94”) = 0.0360. Fig. 3 shows the diffraction pattern of (201) reflection for HoFeG3 at a scattering angle 28 = 67” and a temperature T = 80 K. The (201) peak area was evaluated from the difference of the spectra taken on the crystal set in Bragg position and when rotated by 1.5” out of it. The analysis of the

IN YFeO, AND HoFeO,

537

scattering conditions using a scattering diagram analogous to that one presented in Fig. 2 makes sure that in this case too the contribution of other crystallographic planes to the recorded diffraction intensity is eliminated. The data treatment in this case results in an estimate of the effective spin component Sb = 0.0268 for HoFeG3. Finally, if one assumes the effective spin to be equal to the ideal spin value of Fe3’-ion SW = 512, the mean values of the ratio h/G. are estimated with an accuracy of about 10 % as 0.014 for YFeG3 and 0.0107 for HoFeG3.

References

1. G. Gorodetsky and D. Tteves Whys. Rev. 135A 97 (1964)

4. T. Yamaguchi .I Phys. Chem. Solids 35 479 (1974)

2. V Plakhtii, Yu. Chernenkov, J. Schweizer and M. Bedrizova JETP (Russia) 80 No 6 2465 (198 1)

5. A. Moskvin and E. Sinitsin fiz. Tveni. Tela. (Russia) 17, No 8 2495 (1985)

3. A. Moskvin and E. Sinitsin Fiz. Tved. Tela. (Russia) 14, No 9 2535 (1972)

6. L. Ananiev et al., JINR Commun. Rl3-89-517, Dubna, 1989; also Yu. Taran (ed) Neutron experimental facilities at JINR User k Guide (1992)